mastra-jev-moderation

What it does

Add fail-closed input moderation to a Mastra agent with one Jev request and a typed tripwire instead of parsing a language-model verdict.

How you can use it

A developer can add this to a Mastra chat agent to review messages before answering. They copy the provided moderation file into the project. The app connects to Jev, an outside AI service, using an access key. Jev checks the latest user text for hostile content.

The developer can write plain English rules to allow safe discussions about normally risky topics. If the tool blocks a message, it stops the agent from replying. If the outside service fails or takes too long, the app lets the message through.